For any sonic slowness log, whether it?s monopole compressional, dipole shear, monopole Stoneley, wireline, logging-while-drilling (LWD, etc., one can always ask the question: ?Are these slownesses correct?? Most of the current processing methods for estimating slowness use time-based coherence processing, which relies on coherence as the key quality control (QC indicator. Time-based coherence methods are excellent and robust, but they do not answer the accuracy question. One method is to overlay the estimated slowness on the dispersion curve (i.e., slowness versus frequency that characterizes the wave propagation across the array of receivers. For dispersive borehole arrivals such as dipole flexural (e.g., leaky compressional, it is the low-frequency limit of the dispersion curve that is the correct formation shear (e.g., compressional slowness. While overlaying the processed estimated slowness on the dispersion curve is a well-defined procedure for identifying correct slowness estimation, it is a single depth-by-single depth analysis that is time-consuming. We have developed a new QC method in log form (slowness versus depth that accurately evaluates the correctness of a sonic log.
The new QC method is called slowness frequency analysis (SFA. In SFA, a dispersion curve is generated at each depth from the recorded waveforms, slowness versus frequency information of the dispersion curve is projected onto the slowness axis, and then the slowness projection at each depth is plotted as a log versus depth. The estimated slowness log from time-based coherence processing is then overlaid. For dipole flexural signals, if the estimated slowness log lies at the lowest limit of the SFA projection, then the estimated slowness matches the low-frequency limit of the dipole flexural signal and the slowness log is correct. It is correct because it is consistent with the dispersion curve that describes the data.
This new QC method applies to all modes of sonic propagation. This paper shows numerous wireline and LWD field examples from monopole and dipole data in both fast and slow formations; in formations with radial slowness gradients (i.e., alteration; and in formations exhibiting anisotropy. From the SFA projection logs, we show examples in which the estimated slowness logs are both correct and incorrect. When results are determined to be inaccurate, the log analyst can reevaluate the processing and identify the proper methods to ensure accurate slowness estimates that are consistent with the dispersion curves.